You do not need this directory to run BREOS — the defaults (locations, costs, emissions, PV modules, the bundled load profile) are packaged inside the installed breos package. This folder exists so you can read those defaults and keep your own runnable example configs.

configs/
├── base/          # editable copies of the packaged JSON presets (reference only)
├── examples/      # runnable CLI configs for `breos run`, `sweep`, and `montecarlo`
└── optimization/  # nested configs for the Python optimization API

• base/ mirrors the packaged presets (locations, costs, emissions, financials, electricity). Read them to see what BREOS ships and to copy values into your run config. The CLI always loads its own packaged copies, so editing files here is for reference — it does not change a run.
• examples/ holds CLI configs: single-run inputs for breos run, plus dedicated sweep and Monte Carlo examples. Every file here validates with breos validate-config.
• optimization/ holds configs for the optimization API. These use a different, nested shape and you load them from Python, so breos run and breos validate-config reject them.

# Run a packaged example
breos run --config configs/examples/quickstart.toml

# Check a config without running it
breos validate-config configs/examples/quickstart.toml

# Override any key on the command line
breos run --config configs/examples/pv-only.toml --battery-kwh 5

# Monte Carlo over weather years + demand (needs a multi-year weather file)
breos montecarlo --config configs/examples/montecarlo.toml --runs 100 --plots

# Parameter grid over a base scenario
breos sweep --config configs/examples/sweep.toml --output sweep_results.csv

breos montecarlo runs the scenario as repeated multi-year projections, resampling a weather year and a demand multiplier for each projection year. It writes one row per run to monte_carlo_results.csv and a provenance JSON with the resolved settings and input/output hashes. Pass --collect-yearly to also write the per-run, per-year energy, degradation, and discounted-cost ledger needed for cost envelopes. Pass --plots to generate payback, NPV, grid-independence, final-SoH, and LCOE distributions in plots/. It needs a multi-year historical weather CSV referenced by the [montecarlo] section — BREOS does not bundle weather data. Drop your file in a local weather/ directory (git-ignored) and see examples/montecarlo.toml.

The established demand multiplier is normal with load_uncertainty as its standard deviation. Set load_distribution = "uniform" to use [1 - load_uncertainty, 1 + load_uncertainty]. Weather-year bounds, energy-conserving hourly-to-15-minute interpolation, and worker count are also explicit [montecarlo] settings.

The catalogue keys used in a config (location, pv_module, cost_preset, emissions_country, load_profile) come from the packaged presets. List the valid values with:

breos list locations
breos list modules
breos list cost-presets
breos list emissions
breos list load-profiles

Start from pv-plus-battery.toml if you want to see the full set of knobs; copy any example and edit it for your own scenario.

breos run picks one design and simulates it. The optimization API searches for a design instead, and it takes a nested config that the CLI does not accept.

Load it from Python and pass it to breos.optimization.optimize_system_multi_objective. Needs the pymoo extra (pip install "breos[optimization]"). The full walkthrough is in the Optimization guide.

• Keep public examples on the bundled load_profile = "demandlib_h0" (canonical key "1") unless the example explicitly documents an external, user-licensed RLP directory.
• For external RLPs, use examples/external-rlp.toml as a template and put the licensed CSV files in a local directory such as external_rlp/ (do not commit third-party RLPs).
• breos run configs are mostly flat key/value files (TOML or JSON). [[pv_arrays]] describes multiple arrays and [costs] holds explicit cost overrides. The [sweep] and [montecarlo] tables are read by their dedicated CLI commands; sweep entries can use quoted dotted keys such as "costs.electricity_cost".
• Configs written for the research pvbat engine — with nested model sections, inheritance, or simulation-type blocks — are not compatible. BREOS rejects unknown top-level keys rather than silently applying defaults. Translate the values you need into the flat keys shown in pv-plus-battery.toml (and [montecarlo] for MC studies).